280 ALASKAN GLACIER STUDIES
applied. Many of them were merely due to the cracking of the thin, grassy cover of the till, and across quite a number still extended fine threads of grass roots which had been drawn out as the turf on either side was stretched and broken. The whole arrangement was characteristic of an expanding mass whose turf cover fitted it before the strain was applied but which had since bulged upward so that the turf covering was too small.
Over the top of the 24-foot hillock which dammed back the lake ran the abandoned channel (a, Fig. 84) of the marginal stream of 1909, its site marked by rounded stream gravels and by the absence of vegetation. In late June and early July, 1910, the stream course at the outlet of the lake was at the outer edge of the till mass where it was being pushed up against the mountain side.
It was also evident that there had been several minor lake stages preceding the current one, for on the lakeward slopes of the 24-foot hillock were minute, abandoned shorelines. They were not horizontal, however, but inclined.
The continuation of this crowding forward of the till dam resulted in several changes in drainage and topography along this part of the glacier margin between July 2 and September 6, 1910. Several of the minor lobes advanced 20 to 40 feet and there was much more fracturing of the turf than in July, the turf was being pushed out into the river just below the falls where there was only a mud shore in July, the course of the lake outlet was changed, the ten-foot waterfall of early summer being abandoned, the lake level had been lowered slightly by the downcutting of the new outlet which had two channels instead of one. These channels had cut away a large part of the 24-foot hillock, destroying portions of the abandoned 1909 stream course, and the new waterfalls were only 5 or 6 feet high.
The site of Gilbert's 1899 photograph of the glacier margin described on p. 276 was immediately north of the marginal lake just described. In June, 1910, the ice had advanced clear to the 1892 maximum, several of the prostrate trees killed at the earlier date being still recognizable; but by September they had been overridden. Along this portion of the glacier margin the windrows of dead trees killed in 1892 were still undisturbed in August, 1909. They lay in regular piles, generally with their tops pointing away from the glacier. In June, 1910, the ice had advanced up to many of them and they were heaped up in irregular masses, forming a tangle of wood, turf, and rock debris in one place 75 or 100 feet high.
The changes along this margin of the glacier will not be described in detail, although shown in several photographs. It is enough to say that there were great changes from August, 1909, to June, 1910, and from then to September, 1910. The changes varied, depending on whether the ice was ploughing up turf or forest, and whether it was advancing up a hillslope or down a declivity. All of the barren zone which Gilbert described in 1899, was overridden in 1910, so that forest not touched in 1892 was then being destroyed all along the margin, excepting possibly on a terminal moraine described later.
There were two lower lakes along the eastern margin of Columbia Glacier. In 1909 the lower or westernmost and smaller of these extended eastward from a low waterfall over a rock ledge a quarter of a mile east of the till dam pushed up in 1910. Below this there was no lake in 1909, and as already stated, tidewater then extended up to these falls, three quarters of a mile from Columbia Bay. The till dam of 1910 made a new lake, both longer and deeper than the older one, covering the site of it and of the low water-